HYPOTHESIS TESTING I. Introduction. A. The role of hypothesis testing in the scientific endeavor. B. The logic of hypothesis testing. 1. Begin with a hypothesis based on theory or previous research. 2. Collect empirical data to test the hypothesis. 3. Analyze the data to see if hypothesis is correct. C. The relationship between estimation and hypothesis testing. II. The steps in hypothesis testing. A. State your null hypothesis. 1. A statitistical test must have something to test or evaluate. That something is the null hypothesis. It is typically a simple mathematical or verbal statement about population parameters such as "the two variables are independent" or "the two variables are equal." 2. The null hypothesis is always stated in the null or most neutral form. Typically, this means to state that two values are equal or that there is no difference in two values or that there is no relationship between two variables. 3. The null may not be what the researcher actually hypothesizes to be true. The researcher may actually hypothesize that there is a difference or that two values are not equal. This alternative hypothesis is called the research hypothesis. 4. The null is tested because it is easier to evaluate. 5. If the null is rejected it offers support for the research hypothesis. B. Determine the best statistical test. 1. Each hypothesis test requires that a test statistic be calculated. The test statistic calculated depends on the type of hypothesis being evaluated. 2. Underlying each test statistic is sampling distribution by which the probability of obtaining certain test statistics can be evaluated. 3. There are several different sampling distributions. Thus far we have talked about two: the z and t distributions. Others will be introduced later. C. Check the basic assumptions. 1. Every test statitistic and sampling distribution is based on certain assumptions. Some of the most common assumptions relate to the nature of the sample (is it a probability sample), the level of measurement of the variable being analyzed (is it nominal, ordinal, or interval), and the shape of the sampling distribution (is it normal). However, there are other assumptions as well. 2. Before performing a statististical test, you must be certain the assumption on which it is based hold true. D. Select an alpha level (significance level). 1. Alpha level, sampling error, and probability. 2. Alpha level as the probability of rejecting a null hypothesis when it is actually true. 3. One- and two-tailed tests. E. Calculate the test statistic. F. Determine the probability associated with the test statistic. G. Compare the obtained probability to the alpha level and make a decison about the null.